THE CONSTRUCTION AND USE OF PSYCHOLOGICAL TESTS AND MEASURES
Application of Fuzzy System in Psychological Tests ... · Application of Fuzzy System in...
Transcript of Application of Fuzzy System in Psychological Tests ... · Application of Fuzzy System in...
51
IPA International Journal of Psychology Vol. 9, No. 2, Summer & Fall 2015
PP. 51-75
Iranian Psychological
Association
Application of Fuzzy System in Psychological Tests: Optimize the Number of Questions
for WHOQOL-BREF
Hamid Reza Ghanbari* Department of psychology
University of Sistan and Baluchestan
Mehrdad Mazaheri, PhD Department of psychology
Ferdowsi University of Mashhad University of Sistan and Baluchestan
Hasan Mishmast nehi, PhD Department of mathematics
University of Sistan and Baluchestan
The WHOQOL-BREF is one of the best known questionnaires for measuring the quality of life. It is currently available in more than 40 languages and has been used frequently in cross-cultural comparison studies on subjective well-being and quality of life studies. Some research shows that due to certaincultural biases Iranian respondentshave no tendency or willingness to responding some questions, more specially, question 21, in WHOQOL-BREF. The main aim of the current research was to use the ability and flexibility of fuzzy systems to analyse the WHOQOL-BREF questions to reduce some unclear or doubtful questions. A fuzzy system model proposes to diminish the errors that produce ambiguous concepts by not responding to certain biased questions.The WHOQOL-BREEquestionnaire was analyzed and both the traditional model and the fuzzy model analyses were compared for results using fuzzy systems.As a result, question 21 was removed from the WHOQOL-BREF questionnaire used by the fuzzy system. Keyword: WHOQOL-BREF, psychological test, fuzzy system, Wang & Mandel method, Quality Of Life, Cultural Bias
The theoretical definitions of the “quality of life” and some
related concepts such as “well–being”, “ happiness”, “ life
satisfaction” and “good life” have preoccupied a wide range of
52
disciplines, dating as far back asAristotle (384-19 BC) and early
Greek philosophy (Bowling, 2001). Aristotle, using the Greek
eudaimonia ( a concept which for Aristotle meant having an
understanding of the best way to live one’s life), whichis
commonly translated as ‘happiness’, affirms that the quality of
life is highly relative: it means different things to different
people, and conditions for happiness vary according to a
person’s current condition. Happiness for Aristotle was the
product of activities directed towards clearly defined goals
which inform our whole life rather than being simply short-term
(Chang, Killingworth, & Noaln., 1997).
Self-report rating scales are one of the most common
methods to assess overall SWB. Sandvik, Diener, and Seidlitz
(1993) suggest that standard self-report measures of SWB are
adequate for most research as there is “a unitary core of
experience for well-being, which self-reports reflect to a great
extent. Thus, researchers using standard well-being scales can
generally expect to obtain meaningful, interpretable information
from these scales under ordinary conditions” (p.337). This is
partly attributable to the moderate stability of SWB across
situations and over the life span (Diener & Lucas, 2000). For
example, SWB has been found to correlate 0.85 over a four-year
period (Diener, Suh, & Oishi, 1997).
According to Andrich (1978) and Guilford (1954), there are
four main reasons for the frequent use of the rating scale method
of measurement:
1. Rating scales are relatively easy for researchers to
construct and use compared with other scale formats.
2. They provide the respondent with a limited number of
response options, facilitating data registration for both the
subject and the researcher.
3. The numbering continuum provides respondents with a
53
ruler upon which they can mentally gauge the intensity and/or
direction of their reactions to a statement.
4. Accuracy and reliability of one’s ability to communicate
mental maps increase with experience in using mental rulers.
Also, conceivably, the repeated use of similar formats increases
the accuracy and reliability of the measurement process. One
internalized ruler can be used to measure directions and
intensities across a variety of sentiments.
There are four major methods for constructing rating scales:
Thurstone type or equal-appearing interval scales (Thurstone &
Chave, 1929), Guttman type or cumulative scales (Guttman,
1950),Semantic differential scales (Osgood, Suci &
Tannenbaum, 1957), and Likert-type orsummated rating scales
(Likert, 1932).
The Likert-type rating scale was recently proclaimed to be
one of the most important tools in attitude and survey
measurement (Bergstrom & lunz, 1998).The use of this
measuring device in psychological and educational settings is
virtually universal. Attitudinal data in marketing and public
opinion research and many types of organizational surveys rely
heavily on the rating scale method of measurement (Green &
Rao, 1970). On a Likert or Likert-type scale generally associated
with the work of Likert (1932), respondents are presented with a
series of statements and they are asked to indicate their degree
of agreement (or disagreement) with each item. Responses are
usually made on a 5- or 7-point scale, with response categories
ranging from “strongly agree” to “strongly disagree”, or other
scale point labels referring to frequency or quality. All items are
considered to be of equal value, and response to an item is
weighed to reflect the degree of agreement or disagreement. The
scale score may be either the total number of points (over all
items) or the average of all the item scores. Since the total scale
54
score is obtained by adding the scores from individual items, a
Likert scale is also referred to as a summative scale.
In light of the extensive use of rating scales, it would be also
useful to have a clear understanding of how to optimize
reliability and validity through use of the scale. The number of
rating points used on the “ruler” can vary from 2 to 100 or more.
There is a general consensus that the optimal number is from 5
to 7 points. However, the specific number of points eludes
researchers. Whether to use an even or odd number of categories
is another source of debate among practitioners (e.g., Gable &
Wolf, 1993). Cox (1980; p.408) provides the following
definition of the optimal number of rating points: “At a general
level, a scale with the optimal number of response alternatives is
refined enough to be capable of transmitting most of the
information available from respondents without being so refined
that it simply encourages response error. At that optimal
number, the ratio of meaningful or systematic variation on total
variation is maximized. At an operational level, the optimal
number depends on the purpose of the scale and, thus, the nature
of its systematic variation.”
The WHOQOL-BREF is one of the best known instruments
that has been developed for cross-cultural comparisons of
quality of life and currently it is available in more than 40
languages. The WHOQOL-BREF is a modified version of The
World Health Organization Quality of Life Instrument. The
abbreviated version contains 26 questions divided into four
domains. The WHOQOL-BREF has 26 items derived from the
WHOQOL-100. The items are rated on a 5-point Likert scale.
The four domain scores are scaled in a positive direction, with a
score range of 0-20, and with higher scores denoting higher
QOL. It also includes one facet ofoverall quality of life and
general health. These items are scaled in a positive direction,
55
with a score range of 1-5, and with higher scores denoting a
better quality of life and general health.
(Fig. 1)
Figure 1
WHOQOL-BREF Questionnaire Chart
Cultural biases have been shown not to suffice to explain the
major differences in SWB between countries (Veenhoven,
56
1997), but they still pose a major problem to the international
comparison of QOL data (Schimmack et al., 2002). Asians, for
instance, usually have intermediate mean ratings, with a much
narrower distribution than Westerners (Diener, Smith & Fujita.,
1995). This phenomenon has been attributed to the cultural
valuation of moderation (Diener et al., 1995). Another factor
possibly contributing to the distinctive distribution of SWB
among Asians is less individualism and the persistent closeness
of family ties. Such ways of life are characterized by increased
control, and thus tend to limit the impact of perturbations on
SWB and they affect balance, whether by buffering negative
events or by blunting positive affects.
According to their personality, living circumstances or
culture, people may vary in response to questions concerning
overall life (dis)satisfactions. Schwarz and Strack (1999) have
identified several sources of bias in (conventional) self-reported
global assessments of SWB. These include assimilation and
contrast effects, when current feelings are coloured or
discoloured by the past. Similarly, the mood of the day or events
of the moment may distort responses. In addition, responses
about well-being may be biased by social acceptability (Schwarz
and Strack, 1999).
Iranian people don’t have any tendency toward answering
some questions, because the questions are related to their sexual
relationships.
As the main aim of the current study, we were interested in
using the ability and flexibility of fuzzy systems in order to
eliminate some inappropriate questions from WHOQOL-BREF
questionnaire. .A fuzzy system was designedto diminish the
errors that causeambiguous concepts by not answering specific
questions.
Question 21 was removed from thequestionnaire by using the
57
fuzzy system.We aimed to calculate this question from the
measure of the social relation dimension of the questionnaire
(DOM-3) by removing this specificallyculturally biased
question. (Fig 2)
Figure 2
The Social Dimensionof the WHOQOL-BREF
Questionnaire Chart
Fuzzy Systems and the Features:
The conclusions of psychological studies and educational
sciences are based on uncertain data at this stage and they
resulted in inaccurate fuzzy concepts (Zetenyi, 1998).
Fuzzy systems are the systems based on knowledge or rules;
they are very suitable tools for the modeling of uncertainties and
ambiguities on the basis of the Fuzzy Sets Theory (Wang,
1997).
A fuzzy inference system is a set of fuzzy input(s), fuzzy
rules and fuzzy output(s) that can receive input data accurately
and give the final output in the form of accurate numbers.
Inputs of the fuzzy inference system are the concepts that
psychological and educational researches use to explain the
output(s) (Smithson & Verkuilen, 2006).
The Deduction Process and Inference Process of the fuzzy
system have five stages:
58
Fuzzification, Application of Fuzzy Operators, Conclusion,
Composition, and Defuzzification
1-Fuzzification
In this phase, we putaccurate data into the system and
determine the measure of their attachment to fuzzy sets by
membership functions. In fact, we measure the satisfaction of
(the IF) part of the rule.
2- Application of Fuzzy Operators:
Where (the IF) part is made of some statements, we use the
fuzzy operators to compose statements and determine the
conclusion of (the IF) part of the rule.
3-Inference:
At this point, we define (the THEN) part on the base of (the IF)
part. The input of inference is a number from the prior stage and
the output is a fuzzy set. This step is applied once toeach rule.
4- Composition:
At this stage, we compose different outputs of rules to each
other, (each of them is a fuzzy set) and it determines the overall
space of the fuzzy system. This stage is used once for each
output variable.
5- Defuzzification:
We estimate certainty from uncertainty at this stage. The
input of the defuzzification stage is a composed fuzzy set and
the outputof the defuzzification stage is an exact number. A
fuzzy rule consists of a set of (IF-THEN) fuzzy rules. It is the
heart of the system, because we use other implements of the
system effectively to obtain the rules.
59
To analyze the psychological and social phenomena is one of
the most important abilities of the fuzzy system, the possibility
of quantitative analysis based on fuzzy rules for data that are not
exact (Smithson & Verkuilen, 2006).
In this research, we design an empirical and scientific rules
based on the capabilities of thefuzzy system with the search
table of (Wang & Mandel) input - output data which has three
input parameters: personal relations relationships, sexual
relationships, and social supports of the subjects.
We obtain these three input parameters on the basis of the
measure of psychological tests (WHOQOL-BREF), and estimate
the fuzzy output parameter from the view of social relationships,
and then comparethis conclusion with the conclusion of the
normal method.
Method
A sample of 130 voluntary participants including 30 Swedish
residents, 30 Iranians living in Sweden, and 80 Iranians living in
Iran were asked to fill out the WHOQOL-BREF questionnaire.
The MATLAB software was applied to design the fuzzy
research system by (Wang – Mandel) input-outputsearch table.
We design fuzzy systems to analyze complex psychological
tests by fuzzy (linguistic) rules and nervous networks, because
of the fuzzy rules are an effective and convenient approach
tothepairs of input-output data. The Wang-Mandel approach is
composed offuzzy (linguistic) rules and the nervous networks in
psychological tests and it makes an appropriate method fromthe
pairs of input-output data. It has five basic steps (Wang, 1997).
We design the fuzzy membership function of variables of
personal relations relationships, sexualyrelationships, social
supports of the subjects in terms of the WHOQOL-BREF
questionnaire with 90 Swedes, Iranians living in Sweden, and
60
Iranians living in Iran, toestablish the questionnaire by fuzzy
sets in the form below (fig 3)( Klir,& Yuan, 1995).
61
Figure 3
Membership Functions of Fuzzy Variables in the
Domainof Social Relationships (Personal Relations,
Sexualy Relations, Social Support)
Five basic steps of the Wang-Mendel method are (Wang, 1997):
1- to determine the input-output membership function and to
determine membership degree of each data
2- To attain a rule for each pair of input - output data.
3- To determine a weight and a degree for each rule. We obtain the
degree of each rule by determineof the membership degree of
components and the membership degrees of the pairs of data.
4- To choose the rule that has the maximum degree, we obtain the
number of fuzzy rules in return for the different reasonsAt this
stage we might have similar introduction of (the IF) part and
different conclusion of (the THEN) part, so we have a conflict.
In this case, to resolve the conflict, we choose the rules that have
the highest weight and degree.
5- To determine the measure of the output based on the pairs of
input data related formulas (defuzzification formulas).
62
RuleBases
We use the Wang & Mendel method to design the fuzzy systems
of the research. We consider 37 rules for the fuzzy system; two
of them are mentioned below:
1. If you are very satisfied with your personal relationships, and
very satisfied with your sex relationships, and very satisfied
with your social support, then your social relationship is very
high (weight of rule is 1).
2. If you are very satisfied with your personal relationships, and
very satisfied with your sex relationships, and satisfied with
social support, then your social relationship is very high (weight
of rule is.66).
3. If you are satisfied with your personal relationships, and very
satisfied with your sex relationships, and very satisfied with
social support, then your social relationship is very high (weight
of rule is.66).
4. If you are very satisfied with your personal relationships, and
satisfied with your sex relationships, and satisfied with social
support, then your social relationship is high (weight of rule
is.66).
5. If you are satisfied with your personal relationships, and very
satisfied with your sex relationships, and satisfied with social
support, then your social relationship is high (weight of rule
is.66).
6. If you are very satisfied with your personal relationships, and
very satisfiedwith your sex relationships, and neither satisfied
not dissatisfied with social support, then your social relationship
is high (weight of rule is .66).
7. If you are satisfied with your personal relationships, and
satisfiedwith your sex relationships, and very satisfied with
social support, then your social relationship is high (weight of
63
rule is .66).
8. If you arevery satisfied with your personal relationships, and
very satisfied with your sex relationships, and neither satisfied
not dissatisfied with social support, then your social relationship
is high (weight of rule is .66).
9. If you are very satisfied with your personal relationships, and
satisfied with your sex relationships, and neither satisfied nor
dissatisfied with social support, then your social relationship is
high (weight of rule is 1).
10. If you are satisfied with your personal relationships, and
satisfiedwith your sex relationships, and satisfied with social
support, then your social relationship is high (weight of rule is
1).
11. If you are satisfied with your personal relationships, and very
satisfied with your sex relationships, and neither satisfied nor
dissatisfied with social support, then your social relationship is
high (weight of rule is 1).
12. If you are very satisfied with your personal relationships, and
very satisfied with your sex relationships, and dissatisfied with
social support, then your social relationship is high (weight of
rule is 1).
13. If you are neither satisfied nor dissatisfied with your personal
relationships, and satisfied with your sex relationships, and
satisfied with social support, then your social relationship is
high (weight of rule is 1).
14. If you are neither satisfied nor dissatisfied with your personal
relationships, and very satisfiedwith your sex relationships, and
satisfied with social support, then your social relationship is
high (weight of rule is 1).
15. If you are satisfied with your personal relationships, and
satisfiedwith your sex relationships, and neither satisfied
nordissatisfied with social support, then your social relationship
64
is high (weight of rule is.66).
16. If you are satisfied with your personal relationships, and neither
satisfied nor dissatisfied with your sex relationships, and
satisfied with social support, then your social relationship is
high (weight of rule is .66).
17. If you are satisfied with your personal relationships, and very
satisfied with your sex relationships, and dissatisfied with social
support, then your social relationship is high (weight of rule
is.66).
18. If you are very satisfied with your personal relationships, and
neither satisfied nor dissatisfied with your sex relationships,
and neither satisfied nor dissatisfied with social support, then
your social relationship is high (weight of rule is.66).
19. If you are neither satisfied nor dissatisfied with your personal
relationships, and very satisfiedwith your sex relationships, and
neither satisfied nor dissatisfied with social support, then your
social relationship is high (weight of rule is.66).
20. If you are neither satisfied nor dissatisfied with your personal
relationships, and satisfied with your sex relationships, and
satisfied with social support, then your social relationship is
high (weight of rule is.66).
21. If you are neither satisfied nor dissatisfied with your personal
relationships, and neither satisfied nor dissatisfiedwith your sex
relationships, and very satisfied with social support, then your
social relationship is high (weight of rule is.66).
22. If you are issatisfied with your personal relationships, and
satisfiedwith your sex relationships, and neither satisfied nor
dissatisfiedwith social support, then your social relationship is
high (weight of rule is.66).
23. If you are neither satisfied nor dissatisfied with your personal
relationships, and satisfiedwith your sex relationships, and
neither satisfied nor dissatisfied with social support, then your
65
social relationship is medium (weight of rule is .66).
24. If you are satisfied with your personal relationships, and neither
satisfied nor dissatisfied with your sex relationships, and
neither satisfied nor dissatisfied with social support, then your
social relationship is medium (weight of rule is .66).
25. If you are neither satisfied nor dissatisfied with your personal
relationships, and neither satisfied nor dissatisfied with your
sex relationships, and satisfied with social support, then your
social relationship is medium (weight of rule is .66).
26. If you are satisfied with your personal relationships, and
dissatisfiedwith your sex relationships, and satisfied with social
support, then your social relationship is medium (weight of rule
is .66).
27. If you are satisfied with your personal relationships, and
satisfied with your sex relationships, and very dissatisfied with
social support, then your social relationship is medium (weight
of rule is 1).
28. If you areneither satisfied nor dissatisfied with your personal
relationsrelationships, and neither satisfied nor dissatisfied with
your sex relationships, and neither satisfied nor dissatisfied
with social support, then your social relationship is medium
(weight of rule is 1).
29. If you are satisfied with your personal relationships, and neither
satisfied nor dissatisfied with your sex relationships, and
dissatisfied with social support, then your social relationship is
medium (weight of rule is 1).
30. If you are neither satisfied nor dissatisfied with your personal
relationships, and neither satisfied nor dissatisfied with your
sex relationships, and dissatisfied with social support, then your
social relationship is medium (weight of rule is .66).
31. If you are neither satisfied nor dissatisfied with your personal
relationships, and dissatisfied with your sex relationships, and
66
dissatisfied with social support, then your social relationship is
low (weight of rule is.66).
32. If you are dissatisfied with your personal relationships, and
dissatisfied with your sex relationships, and dissatisfied with
social support, then your social relationship is low (weight of
rule is 1).
33. If you are dissatisfied with your personal relationships, and very
dissatisfied with your sex relationships, and dissatisfied with
social support, then your social relationship is low (weight of
rule is .66).
34. If you are very dissatisfied with your personal relationships,
andvery dissatisfied with your sex relationships, and very
dissatisfied with social support, then your social relationship is
very low (weight of rule is 1).
35. If you are very dissatisfied with your personal relationships, and
dissatisfied with your sex relationships, and very dissatisfied
with social support, then your social relationship is very low
(weight of rule is .66).
36. If you are very dissatisfied with your personal relationships, and
very dissatisfied with your sex relationships, and dissatisfied
with social support, then your social relationship is very low
(weight of rule is .66).
37. If you are dissatisfied with your personal relationships, and very
dissatisfied with your sex relationships, andvery dissatisfied
with social support, and then your social relationship is very low
(weight of rule is .66).
Inference
The true value of the hypothesis is one of the rules of
calculation, and the true value of the conclusion is one of the
applied rules. These conclusions are especially for the output
variable of each rule in a fuzzy subset.The rule ofthe MIN
67
Inference or rule of Multiplication Inference are often used as
the Inference rule. In MIN Inference, membership function of
the output is a section of height, we calculate it by the real
degree of (the If) parts of the given rules (rule weight, α). In the
Multiple Inference, we measure and calculate the membership
function of output by the real degree of (the If) parts of rules.
Composition
There are some fuzzy subsets for each output variable; all of
these variables composing together apply as a single fuzzy
subset for each output variable. MAX or SUM isusually used for
the Composition.In MAX Composition, we compose fuzzy
subsets of outputs according to the reasonable maximum point
which is made especially considering Inference rules of all fuzzy
subsets.
In SUM Composition, we compose fuzzy subsets of outputs
according to allof the reasonable points which aremade
especially considering Inference rules of all fuzzy subsets.
DeFuzzifier
Sometimes it is useful to review fuzzy subsets which are the
conclusion of the composition process, but sometimes is
necessary changing the fuzzy measure to the more clear
measure. The defuzzification process does this stage. Two of the
most common techniques for the parameters are the Center of
Gravity and the Maximum Method.In the Center of Gravity
method, we calculate the clarityvalue of the output variable by
finding the measure of the variable from the center of gravity of
the membership function of the fuzzy value.
In the Maximum method, we calculate the clarityvalue of the
output variable by finding the measure of a variable which has
the fuzzy subsets with the maximum value of clarity.
68
In this research, we use the product method for Inference,
SUM method for composition, and the Center of Gravity
method for DeFuzzifier.
Mathematical Calculation of Example
We design a fuzzy set in the fuzzy system, and we calculate
the measure of the social relationships for this data: Q22=5 and
Q20=2, so we have DOM-3=16.
0.66( 12) / 4 12 16
( ) 0.66( 20) / 4 16 20
0
y y
fuzzy y y y
otherwise
16 20
12 16
0.66( ) ( 12) ( 20) 2.64
4Area f y dy y dy y dy
16 20
2 2
12 16
0.66( ) ( 12 ) ( 20 ) 42.24
4Moment yf y dy y y dy y y dy
( ) 42.2416
2.64( )
yf y dyMomentCentroid
Area f y dy
We obtain similar conclusions for the other inputs with different
outputs of the fuzzy sets [8].
4. Discussion and conclusions:
We demonstrate the obtained conclusions from raw scores with
and without the fuzzy method (normal method) in Table 1:
69
Table 1
Calculation Results areSocialRangeof 50 Subjects (Healthy and
Sick) No Fuzzy Method (Conventional and the Questionnaire
WHOQOL-BREF) and Using Fuzzy Techniques (Output Fuzzy
System Design).
N Gensiat Q20 Q21 Q22
Dom-3
(Social
Relationships)
Without Fuzzy
Dom-3
(Social
Relationships)
With Fuzzy
1 2 5 5 5 20.00 18.70
2 1 5 5 4 18.67 18.60
3 1 4 5 4 17.33 16.00
4 2 5 5 3 17.33 16.00
5 1 4 4 5 17.33 13.90
6 2 5 5 3 17.33 16.00
7 1 5 4 4 17.33 16.00
8 1 4 5 4 17.33 16.00
9 2 5 3 5 17.33 16.00
10 1 4 5 4 17.33 16.00
11 2 4 5 4 17.33 16.00
12 1 4 4 5 17.33 13.90
13 1 4 5 3 16.00 16.00
14 1 5 5 2 16.00 16.00
15 2 3 4 5 16.00 16.00
16 2 4 5 3 16.00 16.00
17 1 4 4 4 16.00 16.00
18 2 4 5 3 16.00 16.00
19 2 3 5 4 16.00 16.00
20 1 4 3 4 14.67 16.00
21 2 4 4 3 14.67 16.00
22 2 4 5 2 14.67 16.00
23 1 4 4 3 14.67 16.00
70
24 2 5 3 3 14.67 16.00
25 1 4 4 3 14.67 16.00
26 1 4 3 4 14.67 16.00
27 2 4 4 3 14.67 16.00
28 2 3 4 4 14.67 16.00
29 2 3 3 5 14.67 16.00
30 1 2 4 5 14.67 16.00
31 2 4 4 3 14.67 16.00
32 1 4 3 4 14.67 16.00
33 2 4 4 3 14.67 16.00
34 1 4 3 4 14.67 16.00
35 2 3 3 5 14.67 16.00
36 1 4 4 3 14.67 16.00
37 2 3 4 3 13.33 12.00
38 1 4 3 3 13.33 12.00
39 2 4 3 3 13.33 12.00
40 1 4 3 3 13.33 12.00
41 2 3 3 4 13.33 12.00
42 2 4 3 3 13.33 12.00
43 2 4 4 1 12.00 12.00
44 2 3 3 3 12.00 12.00
45 1 4 3 2 12.00 12.00
46 1 4 3 2 12.00 12.00
47 1 3 3 2 10.67 12.00
48 1 3 2 2 9.33 8.00
49 2 1 1 2 5.33 5.40
50 2 2 1 1 5.33 5.40
71
Figure 4
Graph comparing the Social Idomain without Using Fuzzy
Techniques (Conventional) Using Fuzzy Technique for 50
Students
Table 6 shows the predictable mean and standard deviation of
scores of the quality of life with a view to social relationships
through the fuzzy method. They are based on two parameters:
personal relationships, and social supports, they have
anoticeable reduction. We compare the means by the related
samples of t-test; they are presented in Table 3.
72
Table 2
Comparison of the Results of a Questionnaire
WHOQOL-BREF Social Domain with and without
Using Fuzzy
Scores N Mean
Std.
Deviation
Social relations domainofthe
WHOQOL-Brief without using
the fuzzymethod
50 14.64 2.80
Social relations domainofthe
WHOQOL-Brief with usingthe
fuzzymethod
50 14.55 2.79
Table3
Ttest forPairedSamples(Dependent)
Sig. (2-tailed) df t
.659 49 .444
We demonstrate that we can remove question number 21
fromthe questionnaire, according to the rules of the Wang-
Mandel method, and we design a special fuzzy system.There is
no meaningful difference between the mean of the scores before
fuzzification and after fuzzification statistically in which 95%
isn’t reliable. It is a powerful system, because it can estimate the
conclusions of the test questions that are predictable even if
some questions are removed.
We show the standard deviation of the scores after
fuzzification is 2.79, and the standard deviation of the scores by
the conventional non-fuzzy method (normal method) is 2.80, so
we have a little reduction from some annoying latent variables
or parameters among the subjects.
The outstanding advantage of this fuzzy method is that the
73
normal method collects irrelevant variables largely, and then it
cannot control predictable scores accurately.
Our analysis shows that the fuzzy system can be inference
reasonably [10]. We predict and conclude the social dimension
in the questionnaire of the fuzzy systems by two input variables
(Table 4) instead of three input variables, without using sexual
scales. This is the unique feature of fuzzy logic.
We use fuzzy logic in research on human behavior because
there are many uncertain data from psychological tests in real
situations.
Uncertainty in related data leads to obtaining inaccurate
concepts. We use fuzzy data instead of raw data, so we are able
to reduce uncertainty. In this research we suggest a new
dimension of the capability of fuzzy systems in educational and
psychological researches. The major advantage of this modeling
is that it predicts psychological parameters by measuring
uncertainty and replacing it with a quantitative measure of
ambiguity.
References
Aristotle. Ethica Nicomachea,-Traditional Bekker pagination
1095 a13-b6.
Bergstrom, B. A., & Lunz, M. E. (1998, April). Rating scale
analysis: Gauging the impact of positively and negatively
worded items. Paper presented at the Annual Meeting of the
American Educational Research Association, San Diego,
California.
Bowling, A. (2001). Measuring disease: a review of disease-
specific quality of life measurement scales, Open University
press.
Chang, M.C., Killingworth, A., & Noaln, P. (1997). A critique
of the concept of quality of life. International Journal of
74
Health Care Quality Assurance10(2), 80-84.
Cox, Eli P. (1980). The Optimal Number of Response
Alternatives for a Scale: A Review. Journal of Marketing
Research, 17, 407-422.
Diener, E., Smith, H., & Fujita, F. (1995). The personality
structure of affect. Journal of Personality and Social
Psychology, 69, 130-141.
Gable, R. K.,& Wolf, M. B. (1993). Instrument development in
the affective domain: Measuring attitudes and values in
corporate and school settings. Boston, MA: Kluwer
Academic Publishers.
Green, P. E., &Rao,V. R. (1970). Rating scales and information
recovery—How many scales and response categories to use.
J. Marketing,34(3), 33–39.
Guilford, J.P. (1954).Psychometric Methods. New York, NY:
McGraw-Hill Co., Inc.
Guttman, L. (1950). The bases for scaleogram analysis. In S.
Stouffer et al. (Eds.). Measurement and Prediction.
Princeton, NJ: Princeton University Press
Klir,G.,& Yuan,B. (1995). Fuzzy Sets and Fuzzy Logic: theory
and applications. New Jersey: Prentice Hall
Klir,G.,& Yuan,B. (1997).Fuzzy Set Theory: Foundations and
Applications. Prentice Hall
Likert, R.A. (1932). A technique for the measurement of
attitudes. Archives of Psychology, 40-52.
Osgood, CE., Suci, J.,& Tannenbaum, PH. (1957). Measurement
of Meaning. Urbana, IL: University of Illinois Press.
Sandvik, E., Diener, E., & Seidilitz, L. (1993). Subjective well-
being: The convergence and stability of self-report and non-
self-report measures. Journal of Personality, 61, 317-342.
75
Schwarz, N. & Strack, F. (1999).Reports on subjective well-
being: judgmental processes and their methodological
implications, in Kahneman, D., Diener, E. and Schwarz, N.
(eds.), Well-Being: The foundation of Hedonic Psychology
(New York: Russell Sage), pp. 61-84.
Smithson,M.,& Verkuilen, J. (2006).Fuzzy Set
Theory:Applications in The Social Scinces . PSAGE
Publications.
Thurstone, L.L.,& Chave, E.J. (1929). The Measurement of
Attitude. Chicago, IL: The University of Chicago Press.
Schimmack, U., P., Radhakrishnan, S., Oishi, V. Dzokoto and
Ahadi, S. (2002). Culture, personality, and subjective well-
being: integrating process models of life-satisfaction,
Journal of Personality and Social Psychology, 82, 582-593.
Veenhoven, R. (1997).Advances in the Understanding of
Happiness, published in French in Revue Quebecoise de
Psychologie, 18, 29-74.
Wang, L. (1997). A Course in Fuzzy Systems and Control.
Prentice-Hall, Upper SaddleRiver, NJ.
Zetenyi, T. (Ed.) (1998). Fuzzy Sets in Psychology. Oxford, UK:
North-Holland.
Received: 3/ 12/ 2013 Revised : 21/ 6/ 2015 Accepted: 5/ 7/ 2015